If reform dividespublic resources and authority among many independent unitsin a transparent way, decentralization will provide an incentive-compatible means of limiting the power of government to threa-ten powerful elites. If these independent units must cooperate toachieve positive-sum outcomes, and this structure of incentivesis replicated within political parties and the public administra-tion, then decentralization will generate competitive incentivescompatible with deepening democracy and strengthening therule of law

Recognizing this forces us to shift our attention away fromthe “decentralization success factors” stressed by so manystudies, a focus that typically blames failure on absent factorssuch as a lack of political will, low bureaucratic capacity, andinsufficient resources. We are led instead to focus on the hier-archical institutions and the structure of incentives in whichdecentralization operates

reduce political instability. Ininstitutionally underdeveloped democracies, elites who fearthe unchecked power of unfriendly governments may resortto violence to protect themselves and their interests

Where democracies are concerned, a political party systemthat balances the power of local and national elites, forcingthem to cooperate in order to win elections at both levels,can help avoid these risks

Argentina shows that subnational actorshave large incentives to distort a federal system in order to ex-tract resources and reduce political competition at the localand national levels. This subverts the accountability of bothsubnational and national politicians, contributing to policyvolatility, a lack of credibility, and the kinds of poor economicand social outcomes that Argentina has suffered in recent dec-ades

Much of the decentralization literature examines what couldbe characterized as fairly simple linear relationships betweendecentralization and policy-relevant outcomes such as educa-tion and health services, public investments, the level of cor-ruption, and national and subnational fiscal deficits. Decadesof research has shown that decentralization both improvesand worsens all of these outcomes, in different countries at dif-ferent times, leaving us unable to draw broader conclusions

Decentralized countries face theintrinsic problem of soft budget constraints because local pol-iticians have strong incentives to overspend and reap the ben-efits themselves, while nationalizing the cost of their behaviorthrough central bailouts.

Limited access is when societyallows only certain groups or individuals to form organiza-tions that receive specific privileges defined by law (i.e., limitedliability). Open access is when any individual or group mayform such an organization. Most developing countries are lim-ited access societies, whereas most developed countries haveopen access

) often claimed that decentralizationin a diverse country would institutionalize social or ethniccleavages, preventing them from fading over time, and wouldalso provide a power and resource base for separatist leaders.Such a risk will be greater for larger, wealthier provinces thatare more likely to be viable as independent countries. Astraightforward remedy, Myerson suggests, is to have smallerprovinces. He suggests a rule of thumb that no provinceshould comprise more than 20% of the national population.In Pakistan, this would imply the breakup of Punjab intotwo or more provinces. Doing so should provide the addi-tional benefit of increasing democratic competitiveness alongthe lines discussed above

But how do we construct democ-racy-enhancing institutions in countries where democracy isfragile or missing? Decentralization, Weingast suggests, is acrucial part of the answer. By decentralizing government,reformers can square this circle by simultaneously building aculture and experience of democratic practice while limitingthe power of central authorities, and hence the threat—realand perceived—to powerful elites.

In the particular context of a het-erogeneous country with spatial concentrations of identitygroups, decentralization can drain political tensions if it cre-ates governments more responsive to the demands and inter-ests of these groups.

Decentralization, it is argued, can relieve thesetensions by handing control over subnational governmentsand resources to local leaders throughout the country, whoin some regions will be leaders of the aggrieved group(s

Decentralization re-orientsthese incentives; this is its most important effect. “Local” offi-cials become local officials, whose tenure and career prospectsare in the hands of the citizens they serve, who elect them.

subnational actorshave large incentives to distort a federal system to their ownends. How can decentralizing countries avoid this

In any country,the goal is to make sure there are at least two strong nationalparties, each of which includes many politicians with goodreputations for public service, and that new parties can devel-op when old parties do not meet voters’ expectations

opportunities for independently elected politicians to demon-strate their abilities to govern

Decentralization further increases competition through sev-eral distinct but related mechanisms

(3) If localities are insufficiently homogeneous, decentralisation may paradoxicallydiminish the accountability of government to interests that are marginalised inlocalities but enjoy some degree of cohesiveness at central level.

‘captured’ by interest groups with preferences for lower taxation than is represen-tative of most of the population. There are two ways in which this process mightbe modelled. First, coalitions of voters across regional boundaries are more likelyto consist of rich than of poor voters. Rich voters are more likely to derive taxableincome from capital that is internationally diversified, so that shocks are correlatedacross regional boundaries; poor voters, by contrast, are more likely to depend onlabour income the shocks to which are more regionally specific. Then the effect ofcentralisation is to increase the influence of rich voters in the overall policy-mak-ing process, and therefore potentially to lower the willingness of the governmentto raise taxes

Fur-thermore, if some politicians are potentially mobile, in the sense that they have asignificant probability of forming the government under either central or localarrangements, their incentives are perverse. They will have an interest in arguingfor centralisation only when its impact in reducing accountability outweighs itsbeneficial impact on intemalising spillovers, since under these circumstances thelevel of effort they have to undertake will be reduced.

The interest of citizens in one or other form of government is not necessarilyshared by their political representatives. Local politicians will lose, and centralpoliticians gain, from centralisation, whatever the benefits to their citizens. Fur-thermore, if some politicians are potentially mobile, in the sense that they have asignificant probability of forming the government under either central or localarrangements, their incentives are perverse. They will have an interest in arguingfor centralisation only when its impact in reducing accountability outweighs itsbeneficial impact on intemalising spillovers, since under these circumstances thelevel of effort they have to undertake will be reduced.

) Centralisation can be achieved without great loss of accountability whengovernments are neither too entrenched (in the sense that localities have a lowprobability in equilibrium of wishing to eject them) nor too insecure (in the sensethat localities have a high probability in equilibrium of wishing to eject them).This is because, if it is unlikely that one locality will want to eject its governmentit is relatively even more unlikely that a decisive number of localities will wish to,so the loss of accountability from centralisation is relatively great. Likewise, if it isunlikely that any one locality will wish to keep its government, it is relatively evenmore unlikely that a decisive number will wish to.

Centralisation can be achieved without great loss of accountability whengovernments are neither too entrenched (in the sense that localities have a lowprobability in equilibrium of wishing to eject them) nor too insecure (in the sensethat localities have a high probability in equilibrium of wishing to eject them)

a central government could concentrate all its effortson pleasing just enough localities to be re-elected, in the knowledge that efforts toplease the remaining regions would be redundant. The value of the accountabilityterm for the remaining regions would be extremely low in spite of the highcorrelation between shocks, simply because it is the correlation between thedifferent localities’ satisfaction levels that counts,

Note that this does not at all the same thing as saying that similarity betweenlocalities strengthens the case for centralisation. If differences between localitiesare incorporated in their different utility functions or the different distributions oftheir locality-specific shocks (variables that are known to populations and govem-ments before any decision are taken), then both local and central governments areentirely capable of setting different values of xi and xi to reflect these differ-ences. Centralisation makes neither easier nor more difficult the local differentia-tion of policy. It is the degree of correlation of shocks (which are not observed bythe populations) that affects the degree of centralisation. To put it another way,what weakens accountability of centralised government to the localities is not therisk that localities will require different policies; it is the risk that localities will bedifferentially satisfied with whatever policies they have. This risk represents a costof centralisation even if localities do not differ in their preferences for governmentpolicy

First, the analysis confirms the basic intuition that the case for centralisation isstrengthened if there are significant spillovers between localities. That is to say,the likelihood that centralisation will increase government effort and consequentlypopulation welfare is increased if spillovers are large and positive

Thirdly, the model shows the possibility of second-best phenomena:the accountability of government to some localities may actually be higher underfull centralisation than under partial decentralisation.

centralised governmentcan engage in local differentiation of policy, but only with respect to variables thatare fully observable by all parties, not for variables that are only partiallyobservable.

In a world in which contracts were complete itwould make no difference to the outcome which party were allocated that right,since the contract could specify what the necessary actions should be as a functionof the relevant circumstances.

First, the improved accountability that results from decentralisationmatters not just because it enables policy to be differentiated according to thepreferences of localities; accountability improves government’s performance evenwhen all localities have the same preferences.

And it is a direct consequence of the fact that eachregion’s welfare is non-verifiable; there is no way to reward or punish a govem-ment according to its performance except by deciding whether or not to re-electit.

educedprobability that a region will be able to choose to elect or reject a government atelection time purely according to its own view of the government’s performance.

Centralisationallows government to reap benefits from the coordination of policies betweenjurisdictions. However, it has a potential cost, namely the diminished accountabil-ity of government to the wishes of any particular region or locality.

The difference between centralised and decentralised government is then a matterof which groups of electors are collectively given the power to decide thegovernment’s re-election. Decentralisation gives control over the policy variablesof a country to a number of different regional or local governments, but grants tothe electors of each region or locality complete power to decide the government’sre-election. Centralisation grants control over all of the country’s policy variablesto a single government, but ensures that regions and localities no longer have theability to determine re-election individually but must do so in concert with otherswhose interests may not coincide with theirs.

citizens cannot take for granted that politicians and bureaucrats will act in theirinterests, and because they cannot write constitutions constraining precisely theway in which they will do so, that they need to allocate power carefully to thoseparties who will, in pursuing their own perceived interests, be most likely tofurther those of the citizens themselve

has the power to take action, because we should presume that agents will takeactions according to their interests. Deciding who should have the power to takecertain actions is therefore a matter of foreseeing which agents will be most likelyto act in the desired way. The allocation of power matters, in short, precisely whenit is not possible to specify in advance precisely how that power should beexercised

The allocation of agents to actions might matter onefficiency grounds because some agents would have a comparative advantage inundertaking certain actions; but it would not affect the actions actually taken.

For its effectiveness in inducing preference revelation,the Tiebout mechanism depends crucially on the mobility of citizens betweenjurisdictions, not upon where or indeed whether citizens may exercise their vote.Indeed, the model appeals to ‘voting with one’s feet’ precisely because normalvoting, which is conceived purely as a means of informing government, fails to beincentive-compatible in the case of public goods and hence has no useful role toplay.

reveal their true preferences forlevels and combinations of public good provision by means of their locationdecisions.

decentralisation may meanthat the power to decide what a policy should be is devolved to some mechanismof local public choice in the regions and localities concerned. In principle, policymay be locally differentiated without any such decentralisation of power.

ocal jurisdictions discretion tocompete in the provision of public goods will tend to increase the efficiency ofallocation of those public goods for which the Tiebout conditions are not fla-grantly violated.

n the first sense, decentralisa-tion may be used to mean that policies are differentiated according to thecircumstances of different regions and localities. The considerable literature on thetopic of ‘fiscal federalism’ sparked by the classic article of Tiebout (1956), hasdealt largely with this sense of decentralisation. Tiebout claimed that, undercertain circumstances, competition between jurisdictions supplying rival combina-tions of local public goods would lead to an efficient supply of such goods.

ircumstances where governments may act as Leviathansappropriating resources for their own use.

Centralisation allows benefits from policy coordination but has costs in terms ofdiminished accountability, which can be precisely defined as the reduced probability thatthe welfare of a given region can determine the re-election of the government

incomplete contracts.

TaxNegative suggests that a one-percentage-point increase inthe tax rate for this type of fiscal policy—that is, a one-percentage-pointincrease in distortionary taxes accompanied by a one-percentage-pointincrease in either unproductive spending or nondistortionary taxes (seetable 3)—would lower annual economic growth by 0.18 percentagepoints. Alternatively, a one-percentage-point increase in taxes as part of aTaxAmbiguous fiscal package is associated with lower annual economicgrowth of 0.11 percent. If part of a TaxPositive fiscal package, the sametax increase is associated with a 0.02 percent increase in economic growth

ositive economic growth (TaxPositive fiscalpolicies): (1) an increase in nondistortionary taxes to fund productive expen-ditures, (2) an increase in nondistortionary taxes accompanied by a decreasein distortionary taxes, and (3) an increase in nondistortionary taxes todecrease the deficit. Every other combination of taxes-spending-deficits ispredicted to have “ambiguous” (or zero) growth effects

With respect to the other goals of our study, we find strong evidence ofpublication bias. The results are consistent with researchers and journalsdiscriminating in favor of negative estimated tax coefficients. While thebias is sufficiently large to be statistically significant, it is not large enoughto be economically significant on average, though it may be important inindividual studies

We estimate that a 3.5-percentage-pointincrease in taxes (roughly equivalent to a 10 percent increase) as partof a TaxNegative fiscal package is associated with decreased annualGDP growth of approximately 0.2 percent. The same increase in taxesas part of a TaxPositive fiscal package is associated with an increase inannual GDP growth of 0.2 percent. This compares with an averageannual GDP growth rate of 2.5 percent for the countries in our sample.Stated differently, we estimate there is a 0.4 percent difference inannual GDP growth depending on whether the 3.5-percentage-pointincrease in taxes is part of a TaxNegative or a TaxPositive fiscal pack-age. These estimates suggest that taxes, and how taxes are raised andspent, have moderate impacts on economic growth

ble 5. Summary Statistics of Study Characteristi

Our estimation strategy is built around the following specification:^a1i ¼ b0 þ b1TaxNegative i þ b2TaxPositive i þ Xkdk X ki þ ei; ð3Þwhere the dependent variable, ^a1i, is the estimated tax effect from a regres-sion in study i; TaxNegative and TaxPositive are dummy variables indicat-ing that the estimated tax coefficient is part of a tax-spending-deficitspecification where taxes are predicted to have negative and positiveimpacts on economic growth; and the X k are data, estimation, and studycharacteristics that may influence the sizes of the estimated tax effects.The coefficients b1 and b2 can be interpreted as the estimated growtheffects of a one-percentage-point increase in taxes associated with a Tax-Negative and TaxPositive fiscal policy compared to a one-percentage-pointincrease in taxes associated with a TaxAmbiguous fiscal policy

gure 1. Gemmell, Kneller, and Sanz (2009)’s classification of the growth effects ofvarious fiscal policies’ combinations. Source: Gemmell, Kneller, and Sanz (2009,Table 1, page 19

s.Study Colombier (2009)Romero-A ́ vila andStrauch (2008)Bleaney, Gemmell, andKneller (2001)Afonso and Fuceri(2010)Muinelo-Gallo and Roca-Sagal ́es (2013)Regression Table 1, Column 1 Table 5, Column 5 Table 1, Column 1 Table 5, Column 1 Table 3, Column 1DependentvariableReal GDP percapita growthReal GDP per capitagrowthReal GDP per capitagrowthReal GDP percapita growthReal GDP per capitagrowthTax variable(s) - Average tax rate - Direct taxes- Indirect taxes- Social contributions- Distortionary taxes - Direct taxes- Indirect taxes- Socialcontributions- Direct taxes- Indirect taxesOther fiscalvariable(s)- None - Governmentconsumption- Governmenttransfers- Governmentinvestment- Other revenues- Other expenditures- Budget surplus- Productive expenditures- Nonproductiveexpenditures-None - NondistributiveexpendituresOmitted fiscalcategories- All expenditures- Deficit- Deficit - Nondistortionary taxes - All expenditures- Deficit- Distributive expenditures- DeficitCountries 21 OECDcountries15 EU countries 22 OECD countries 28 OECDcountries21 OECD countriesTime period 1970–2001 1960–2001 1970–1995 1970–2004 1972–2006N

This is where meta-analysis can be helpful. By combining estimated taxeffects from many different studies and controlling for the associated tax-spending-deficit specifications employed by those studies, it makes it pos-sible for estimated tax effects to be more closely matched to actual fiscalpolicies. This could be done by estimating a meta-regression equation withthe estimated tax effects as the dependent variable and a series of dummyvariables as explanatory variables identifying different tax-spending-deficitcombinations. This would allow the policy maker to calculate a predictedtax effect for the specific policy being considered. The challenge is howbest to do th

rst, defenseprograms are not efficient ways to create employment. Thus, it is not wise forstates to use defense spending to create jobs for the purpose of stimulatingtheir economies. Second, increased defense spending does not help or hurtstate economies; its impact is minimal. That said, policymakers need tounderstand that defense increases have an indirect harmful effect on privateinvestment. But since the negative impact of defense spending on privateinvestment mostly occurs when defense programs are financed by deficits, itis not necessary to take into account the impact of defense increases oninvestment in policy decisions as long as increases in defense spending arefinanced by means other than budget deficits

efensespending does dampen private investment at the global level, making this acost of enhancing security

Finally, defense spending shows an immediate, negative, and significanteffect on economic growth, although the effect is rather small. A 1% increasein defense spending as a percentage of GDP decreases economic growth by0.06%. Moreover, this negative effect is mostly offset by the positive effects inthe following two years by 0.03% and 0.01% respectively. In other words, a1% increase in defense spending as a share of GDP results in a 0.02% declinein economic growth. This means the direct effects of defense spending oneconomic growth seem to be almost negligibl

. In other words, the effects of military spending on unemployment arelimited, and military spending does not seem to be a good way to reduceunemployme

In contrast to defense spending, nonmilitary government spendingdecreases unemployment with the negative effects occurring in the sameyear and two years later. The coefficients are –2.28 and –1.05 respectively.A large portion of this effect is cancelled out by the positive effect of militaryspending on unemployment in the following year with the coefficient of 3.41.

Nonmilitary government spending also shows an immediate negative andsignificant effect on private investment, indicated by its negative effect on lag 2.However, a positive and significant effect appears in lag 1. The coefficients forthe first and third year are –1.68 and –0.82 respectively, while the coefficient forthe second year is 3.24. In other words, over a three-year period, a 1% increase innonmilitary government spending as a percentage of GDP will lift privateinvestment by approximately 0.7%. Unlike military spending, nonmilitary gov-ernment spending in the long run shows a stimulating effect on private invest-ment, although the impact is limited. Considering that private investment iscritical for economic growth, nonmilitary government spending indirectly helpseconomic growth through enhancement in private investment in the long run.Turning to other control variables, previous economic growth shows a

. In other words, the overall effects ofdefense spending on private investment, as we hypothesized, are negative.Ceteris paribus, a 1% increase in military spending as a percentage of GDP,on average, will bring down private investment by about 5% over a three-yearperio

SUR is a method used to estimate multiple equations as one big equation toimprove efficiency when there are contemporaneous correlations between theerror terms across equations. For example, if there are N equations, Yi = Xiβi + εiwhere the subscript i refers to the ith equation. These equations can be written asY1Y2...Y n2666437775 ¼X1X2...Xn2666437775 β1β2...βn2666437775 þε1ε2...ε

There are three equations in this study. We can estimate them jointly usingZellner’s (1962) Seemingly Unrelated Regression (SUR) or separatelyemploying Ordinary Least Squares (OLS). According to Zellner, joint estima-tion is asymptotically more efficient if the error terms of the equations arecontemporaneously correlated, although OLS is better if they are indepen-dent from each other (Judge, Griffiths, Hill, Lütkepol, and Lee 1985; Kariya1981; Mintz and Huang 1990, 1991). Thus, we employed the Breusch-PaganTest to find out whether the error terms of equations are contemporaneouslycorrelated. If we cannot reject the null hypothesis that the contemporaneouscorrelation of errors across equations is zero, each equation is estimatedseparately using OLS in order to have efficient estimators. The result of theBreusch-Pagan Test shows that error terms of the equations are contempor-aneously correlated (χ2 ¼ 75:784 with p value of 0.0000). Therefore, SUR isprescribed.

Our data cover almost all sovereign states with annualmilitary budgets. 13 We select 1990 as the starting year for two reasons. First, acomprehensive global examination of the defense-growth relationship hasbeen done during the Cold War years (for example, Heo 1998; Mintz andStevenson 1995; Ward, Penubarti, Cohen, and Lofdahl 1995). Therefore, wewant to focus on the post-Cold War era. The second reason is data avail-ability. Except for OECD members, unemployment data for most countriesunder investigation are not available prior to 1990. We also include post-communist states and former Soviet Republics that were established after1990

“opportunity costs.” Themost common argument in this category is that military spending divertscivilian resources from more productive use

military spending both directly and indirectly helps economic growth byincreasing purchasing power, enhancing aggregate demand, and financingheavy industry, especially armaments

unemploymentit ¼ α2 þ β21 unemploymentit1þ Xnj¼0β22 ln military spendingGDP itjþ Xnk¼0β23 ln nonmilitary spendingGDP it

defense indus-tries create fewer jobs per dollar spent than civilian industrie

Theoretically, defense spending generates jobs directly and indirectly.Direct employment refers to direct hiring of active duty soldiers and civiliansworking for the military.

We expect that military expenditureswill dampen private investment, while nonmilitary spending will encourageinvestment by improving infrastructure and education. Theoretically, capitalstock should be included in the model, but we decided to drop the variabledue to the lack of data availability.6 Our model isln investmentð Þit ¼ α1 þ β11Δ ln GDPð Þit1 þ β12 Inflationitþ Xnj¼0β13 ln military spendingGDP itjþ Xnk¼0β14 ln nonmilitary spendingGDP itkþ e1

Every gun that is made, every warship launched, every rocket fired signifies, in thefinal sense, a theft from those who hunger and are not fed, those who are cold andare not clothed. This world in arms is spending the sweat of its laborers, the geniusof its scientists, and the hopes of its children

The direct and indirect link of the defense-growth nexu

By contrast, the supply-side models focus on the aggregate productionfunction and investigate the direct economic effects of defense spending ongrowth. Thanks to the solid theoretical underpinning of these approaches,studies on the supply-side defense-growth relationship have widely used theFeder-Ram defense-growth model as a theoretical structure (for example,DeRouen and Heo 2001; Heo 1996, 1998, 2000, 2010; Heo and DeRouen1998), although Dunne, Smith, and Willenbockel (2005) have recently criti-cized the theoretical and empirical underpinnings of this approach. Dunneet al. (2005:450) summarize the demand and supply-side effects of defensespending on economic growth as follow

Scholars evaluating the impact of defense spending on growth use bothdemand-side and supply-side models. Demand-side models address thenotion of crowding-out; one source of demand competes for scarce (capital)resources with another source, such as investment or social welfare spending.Thus, dependent variables in demand-side models have included savings,investment, employment, and education or public health expenditures.Unfortunately, demand-side studies use variables with inconsistent modelspecifications due to the lack of a dominant theoretical framework forempirical analysis (for example, Heo 1999; Heo and Eger 2005; Mintz 1989;Mintz and Huang 1990, 1991

Hence it is necessary for a prince wishing to hold his own to know how to do wrong, and to make use of it or not according to necessity. Therefore, putting on one side imaginary things concerning a prince, and discussing those which are real, I say that all men when they are spoken of, and chiefly princes for being more highly placed, are remarkable for some of those qualities which bring them either blame or praise; and thus it is that one is reputed liberal, another miserly, using a Tuscan term (because an avaricious person in our language is still he who desires to possess by robbery, whilst we call one miserly who deprives himself too much of the use of his own); one is reputed generous, one rapacious; one cruel, one compassionate; one faithless, another faithful; one effeminate and cowardly, another bold and brave; one affable, another haughty; one lascivious, another chaste; one sincere, another cunning; one hard, another easy; one grave, another frivolous; one religious, another unbelieving, and the like. And I know that every one will confess that it would be most praiseworthy in a prince to exhibit all the above qualities that are considered good; but because they can neither be entirely possessed nor observed, for human conditions do not permit it, it is necessary for him to be sufficiently prudent that he may know how to avoid the reproach of those vices which would lose him his state; and also to keep himself, if it be possible, from those which would not lose him it; but this not being possible, he may with less hesitation abandon himself to them. And again, he need not make himself uneasy at incurring a reproach for those vices without which the state can only be saved with difficulty, for if everything is considered carefully, it will be found that something which looks like virtue, if followed, would be his ruin; whilst something else, which looks like vice, yet followed brings him security and prosperity.

the critical significance of this phenomenon of uneven growthwhen he promulgated the law of uneven development

The most important factor for the process of international po-litical change is not the static distribution of power in the system(bipolar or multipolar) but the dynamics of power relationshipsover time. It is the differential or uneven growth of power amongstates in a system that encourages efforts by certain states tochange the system in order to enhance their own interests or tomake more secure those interests threatened by their oligopolis-tic rivals. In both bipolar and multipolar structures, changes inrelative power among the principal actors in the system are pre-cursors of international political change

markets, are individualist in origin, spontaneously generated, and unin-tended. In both systems, structures are formed by the coaction of theirunits. Whether those units live, prosper, or die depends on their ownefforts. Both systems are formed and maintained on a principle of self-help that applies to the units. To say that the two realms are structurallysimilar is not to proclaim their identity. Economically, the self-help prin-ciple applies within governmentally contrived limits. Market economiesare hedged about in ways that channel energies constructively. One maythink of pure food-and-drug standards, antitrust laws, securities andexchange regulations, laws against shooting a competitor, and rules for-bidding false claims in advertising. International politics is more nearly arealm in which anything goes. International politics is structurally simi-lar to a market economy insofar as the self-help principle is allowed tooperate in the latter (Waltz, 1979, p. 91

In domestic society, as a resource becomes scarce relative tothe demands of society, the increasing cost of the resourcecreates an incentive for individuals, groups, or the governmentto pay the costs of innovations that will satisfy the unmet de-mand. The most important mechanism for stimulating this incen-tive is the creation and enforcement of new types of propertyrights: A right is conferred on the entrepreneur to enjoy thefinancial rewards of his endeavors (North and Thomas, 1973, p.16). Thus the innovation of the patent system extended the no-tion of property rights to intellectual creations in order to encour-age industrial invention

An increase in some inputs relative to other fixed inputs will, in a givenstate of technology, cause total output to increase; but after a point theextra output resulting from the same additions of extra inputs is likely tobecome less and less. This falling off of extra returns is a consequence ofthe fact that the new "doses" of the varying resources have less and lessof the fixed resources to work with (Samuelson, 1967, p. 26).Or, to put it more succinctly, "the output of any productiveprocess will increase at a decreasing rate if the quantity of onecooperating factor of production is kept constant while that of theothers is increased" (Hirschman, 1971, p. 17). Thus, every factorof production (land, labor, and capital) must increase together (inthe absence of technological advance) if any economy is to es-cape the threat of diminishing returns

When one inquires what these interests or objectivesare, one confronts a long-standing debate between what StanleyHoffmann (1973) called the classiques and the modernes. Theformer, mainly political realists, argue that national security andpower have been in the past and continue to be in the presentthe primary objectives of states. The latter counter that, how-ever true this may have been in the past, attaining domesticeconomic stability and ensuring the welfare of the populace havebecome the foremost objectives of states in the contemporaryworl

he state may be conceived as a coalition of coalitions whoseobjectives and interests result from the powers and bargainingamong the several coalitions composing the larger society andpolitical elite. In the language of Brian Barry (1976, p. 159),collective choice and determination of political objectives arecoalition processes (Cyert and March, 1963, p. 28)

Strictly speaking, states, as such, have no interests, or whateconomists call "utility functions," nor do bureaucracies, interestgroups, or so-called transnational actors, for that matter. Onlyindividuals and individuals joined together into various types ofcoalitions can be said to have interests

. Throughout history, in fact, states and political organiza-tions have varied greatly: tribes, empires, fiefdoms, city-states,etc. The nation-state in historical terms is a rather recent arrival;its success has been due to a peculiar set of historical circum-stances, and there is no guarantee that these conditions will con-tinue into the future

However, the state is the principal actor in that the nature of thestate and the pattern of relations among states are the mostimportant determinants of the character of international relationsat any given momen

If the state is to protect its citizens and their rights, and in theabsence of any higher authority and in a competitive state sys-tem, the state must be "self-regarding" and must look on allother states as potential threats

These internal and external functions of the state and theultimate nature of its authority mean that it is the principal actorin the international system. The state is sovereign in that it mustanswer to no higher authority in the international sphere. Italone defines and protects the rights of individuals and group

Within the territory it encom-passes the state exercises a monopoly of the legitimate use offorce and embodies the idea that everyone in that territory issubject to the same law or set of rules.

The delineation of property rights is necessary if any society isto operate effectively; property rights function by conveying"the right to benefit or harm oneself or others. Harming a com-petitor by producing superior products may be permitted, whileshooting him may not. A man may be permitted to benefit him-self by shooting an intruder but be prohibited from selling belowa price floor" (Demsetz, 1967, p. 347). Thus the nature anddistribution of property rights determine which individuals willbe most benefited and which will pay the most costs with respectto the functioning of different types of social institutions. For thisreason the basic domestic function of the state is to define andprotect the property rights of individuals and groups.

delineation

The absolute mon-arch or contemporary politician has personal objectives he seeksto achieve, the primary one being to maintain himself in offic

Principally through its definition and enforcement ofproperty rights the state protects the welfare of its citizensagainst the actions of other individuals and states and also pro-vides a basis for the resolution of disputes

can provide these "public goods" and overcome the free-riderproblem

The theory of the state that we shall use in this study is thatthe state is "an organization that provides protection and [wel-fare] . . . in return for revenue"

Although resolution of a crisis through peacefuladjustment of the systemic disequilibrium is possible, the princi-pal mechanism of change throughout history has been war, orwhat we shall call hegemonic war (i.e., a war that determineswhich state or states will be dominant and will govern the sys-tem). The peace settlement following such a hegemonic strugglereorders the political, territorial, and other bases of the system.Thus the cycle of change is completed in that hegemonic warand the peace settlement create a new status quo and equilib-rium reflecting the redistribution of power in the system and theother components of the system.

In this book, power refers simply to the military, eco-nomic, and technological capabilities of states.

However, the most destabilizing factoris the tendency in an international system for the powers ofmember states to change at different rates because of political,economic, and technological developments. In time, the differen-tial growth in power of the various states in the system causes afundamental redistribution of power in the system

They do not describe the actual deci-sion processes of statesmen, but as in the case of economic the-ory, actors are assumed to behave as if they were guided by sucha set of cost/benefit calculation

4 Once an equilibrium between the costs and benefits of furtherchange and expansion is reached, the tendency is for the economiccosts of maintaining the status quo to rise faster than the economiccapacity to support the status quo.5 If the disequilibrium in the international system is not resolved,then the system will be changed, and a new equilibrium reflectingthe redistribution of power will be established

A FRAMEWORK FOR UNDERSTANDING INTERNATIONALPOLITICAL CHANGEThe conceptualization of international political change to be pre-sented in this book rests on a set of assumptions regarding thebehavior of states:1 An international system is stable (i.e., in a state of equilibrium) if nostate believes it profitable to attempt to change the system.2 A state will attempt to change the international system if the ex-pected benefits exceed the expected costs (i.e., if there is an ex-pected net gain).3 A state will seek to change the international system through territo-rial, political, and economic expansion until the marginal costs offurther change are equal to or greater than the marginal benefit

e argument of this book is that an international system is estab-lished for the same reason that any social or political system iscreated; actors enter social relations and create social structures inorder to advance particular sets of political, economic, or othertypes of interests.

Internationalrelations continue to be a recurring struggle for wealth andpower among independent actors in a state of anarchy.

. It is not sur-prising that International Relations has tended to be discussed in staticterms, and that stability has tended to be interpreted in terms of themaintenance of the status quo. A dynamic approach to InternationalRelations would immediately confront the analyst with no alternativebut to acknowledge war as the only available mechanism for chang

The facts show that DeNolf spoke both online and face-to-face as a private personabout important public issues that affected him directly and not in his official capacity as anemployee of OSU.

. In Demers, the Ninth Circuit held that “Garcetti does not apply to speech relatedto scholarship or teaching.” 746 F.3d at 406 (internal citation and quotations omitted).

In the alternative, if the Policy is tested at intermediate scrutiny, it meets the same fate as it doesat strict scrutiny. Intermediate scrutiny requires means that are substantially related to animportant interest. United States v. Virginia, 518 U.S. at 524 (1996). Under this test, the stateneeds to advance an “exceedingly persuasive justification” required to sustain its decision toclassify by sex. Id. Whether this is a third tier of intermediate scrutiny or a general description ofthe test itself is unclear. Id. at 596 (Scalia J., Dissenting). What is clear is that such a justificationmust be genuine as opposed to post-hoc. Id., at 533 (majority opinion). I do not doubt that theSchool is sincere in its intentions or that it believes that security will be improved. But that doesequate to an exceedingly persuasive justification. For that reason, intermediate scrutiny is notsatisfied. The Policy is unconstitutional. The District Court should be reversed.

Applying strict scrutiny to the case at hand, the Policy fails. OSU lacks a compelling state interest.While the nation and Olympus do have a history of sex discrimination, more of the same will notsolve that problem – one that is in the rear window. We have or have had female judges andjustices, governors, mayors, cabinet secretaries, a speaker of the house (twice), and a vicepresident. Rather than solving a problem, the School has created new problems. Even if the Schoolproffered a compelled interest, which it has not, its means do not advance that interest in a mannerthat is narrowly tailored. Put simply, OSU could have pursued a course of action that was lessrestrictive. There are many ways to attract female students beyond the discriminatory practicesemployed here. The School, for instance, might utilize enhanced recruitment techniques, advertisein female dominated media markets, or aggressively recruit female candidates, to name but afew. Such methods might accomplish similar results. The existence of such alternatives undercutsthe argument that the Policy is least restrictive and thus not narrowly tailored.

But even a program that educates a high number of future police chiefs andfederal law enforcement leadership does not qualify for the service academy exception referencedin Students for Fair Admissions, Inc., 600 U.S. at 214, n. 4 (2023). While such a program may beprestigious and important for law enforcement, it lacks several key characteristics that distinguishmilitary service academies. First, it has no direct connection to national security: While lawenforcement is crucial for public safety, it does not have the same direct link to national defenseas military academies. Second, unlike service academies, criminal justice programs are not underdirect federal control or supervision

hat is more, the School assertsa general need for a more balanced crop of law enforcement leadership, not necessarily a morebalanced student body, but provides no evidence as to why this is compelling or how it improvesthe education of its students let alone law enforcement leadership or the nation’s security.

he correct test for affirmative action cases—including those involving sex classifications—should be strict scrutiny. I am persuaded by JusticeBrennan’s argument in Frontiero v. Richardson that strict scrutiny should be the correct test forsex discrimination. 411 U.S. 677, 682 (1973) (plurality) (finding “implicit support” for theargument that classifications based on sex are “inherently suspect and therefore be subjected toclose judicial scrutiny”). Sexism, and sex discrimination, are wrong – they are stains to beremoved from the larger body politic. But the government must resort to them to achieve itsobjectives. After all, two wrongs do not equal a right.

Most on point is Renkenv. Gregory, 541 F.3d 769 (7th Cir. 2008) (applying Garcetti to a dispute arising from a universityprofessor’s claim that his employer, the University of Wisconsin-Milwaukee, had retaliated againsthim after he criticized the university itself). In its decision, the Seventh Circuit found that Garcettiapplied because “Renken made his complaints regarding the University's use of NSF fundspursuant to his official duties as a University professor.”

In essence, this approach instructs courts todetermine where the speech occurred—namely, did it occur in the workplace—and whether itconcerned a subject matter relevant to the employee’s employment. Garcetti, 547 U.S. 410, 420–21 (2006).

Under cross-examination by attorneys for DeNolf, two witnesses called by the School who wereinvolved in the Policy conceded that had DeNolf been a woman he would have been accepted

2) that Respondent violated his First Amendment rights when it fired him for complaints he madeabout the admissions policy.

(1) that Respondent’s admissions policy thatgives preferential weight to female applicants violated his right to equal protection of the law

alumni donations, and how it would affect OSU’s highly ranked athletic programs, especiallysince recruiting future student-athletes was dependent on these donations.

and large, the persons who contacted OSU were upsetthat the university, by virtue of the fact that it employed DeNolf, appeared to endorse his pointof view.

A great many of these messages came from the parents of students, donors,and alumni. Several elected officials who called upon OSU to take action in defense of the femalegraduate students at the School—many of whom had been “trolled” on social media and had beenlabeled “Affirmative Action Babies!” B

he asserted, “I believe men and women are equal in every way, I simply want everyoneto be treated equally – a level playing field.”

DeNolf, who was very concerned and distraught by events that had occurred on campus, wasexcited about the festival. DeNolf asked several of his colleagues at OSU, including ProfessorBronner, for suggestions as to which of his poems he should recite. DeNolf did so because severalprofessors were familiar with his work. The consensus was that DeNolf should present a poemhe had written about the Policy entitled “Imagine a Level Playing Field.” The poem followed thelyrics and beat of John Lennon’s song entitled “Imagine.” It addressed society’s use ofaffirmative action to supposedly level the playing field for all. The poem’s theme was that totruly create a world that was fair for all society must forgo affirmative action. Before presentingthe poem, which he sang, while playing an acoustic guitar, DeNolf blasted “the powers that be”and asserted that “affirmative action is inconsistent with true equality.” He called for “an end tolaws that favor people based on immutable characteristics” and called for “students to rise up andchallenge inequality whatever form it takes and wherever they find it—including at OSU

At two academic conferences where he presented papers. The first was at aforum on sex discrimination and the law held on OSU’s campus and the secondwas at a professional association meeting held out of state that was funded byOSU travel support. These papers identified DeNolf as a lecturer at OSU

After DeNolf’s application to the School was rejected, he began to complain in class about thePolicy. Specifically, DeNolf complained that his right to equality had been violated. At timeshe called on female students to defend the Policy. DeNolf was critical of men in class who eitherdid not agree with him or were not actively trying to end the Policy. As the term progressed,DeNolf’s attitude became more vociferous, and he began to complain in settings outside hisclassroom. For example, DeNolf voiced his complaints in the following settings

The faculty at OSU is not unionized.

he School does not deny any of the aforementioned. The School contended in the District Court,however, that the female students in question all brought soft variables and experiences, whichallowed them to add to the overall quality of the incoming class in ways in which DeNolf did not.Some of the admitted female students, for example, came from impoverished backgrounds, hadlived abroad, spoke languages that are frequently critical to the success of law enforcementagencies because of the populations they serve, were single parents, had started their ownbusinesses, had been student-athletes in college, and/or had unique artistic talents. DeNolf hadnone of these soft variables or experiences. Thus, while sex was a factor in their admission (and,by extension, in DeNolf’s rejection), the School argues it was not the deciding factor

On March 17, 2023, DeNolf was notified that his application had been rejected. DeNolf’s GREscore of 164 verbal and 166 quantitative put him just above the 75th percentile nationally,6 his 3.5undergraduate GPA placed him in the top 20% of his graduating class at Kedesh College, but justbelow and his 3.75 graduate GPA in graduate school placed him in the top 20% of his graduatingclass at Kedesh College.7 Using the Freedom of Information Act to get data from the School,DeNolf discovered that while every accepted male met or exceeded his portfolio, many of theaccepted female students did not exceed his GPA or his GRE. In fact, female students wereadmitted with undergraduate GPAs as low as 3.25 and with GRE scores as low as 156 in verbaland 158 quantitative—but not necessarily in combination with each other. Many of the successfulfemale applicants had better work experience than DeNolf

Alfie Sasaki, told DeNolf that what he had done was “admirable” butthat “there was no place for nice guys in mock trial and suggested that he try moot court becausethat was where the law students who were not ruthless enough for mock trial went.

6only; offering female applicants who sign up for tours of the School free tickets to athletic eventsor Amazon.com gift cards; offering two season tickets to all female applicants who choose toattend the School (they can choose from the School’s male or female basketball or volleyballteams); creating five faculty lines dedicated to hiring qualified female faculty; creating severalscholarships for qualified female applicants; and offering School hoodies to the first 500 femalestudents who apply. In addition, the School created the Preparation Program for Female CollegeStudents enrolled in OSU’s criminal justice undergraduate program. This program, which is runby OSU Professor Caitlyn Smith, is funded by a $300,000 grant award from the National FemaleLaw Enforcement Professional Association. This program was designed to reverse the trend thatstarted around 2015, as reported in Table 2, by providing participants mentoring, assistance withpreparing for admissions tests, guidance for interviews, and help with personal statements. Despitethese efforts, the gap between male and female enrollment persists, and the School remains anoutlier regarding female matriculation as compared to other schools regionally and nationall

These included raising its recruitment budget by 20%; targetingtalented female students through the School Admissions Council’s Criminal Justice CandidateReferral Service; visiting every police department within 500 miles of the School twice a year;offering 80% of all females who apply fee waivers; hosting application workshops for females

Nor is there is evidence that it used sex as a stereotypeor as a negative when evaluating applicants. Put another way, the School did not treat all womenor men as being monolithic – as representatives for their sex. To the contrary, the School drilleddeep when it considered each applicant looking past demographics. No applicant was awardedpoints or denied points because of demographic factors such as race, sex, ethnicity, religion, orsexual orientation. The Pettus-Cole Policy, which went into effect on July 1, 2022, afforded eachapplicant an individualized review that gave consideration to a range of attributes that would addto the overall makeup of each entering class, that went beyond sex. For example, life experiences,geographical factors, such as where applicants grew up or where they live, languages spoken, ifthey describe themselves as religious, involvement in their community, the background of theirparents, athletic or artistic abilities, and prior relevant vocational experiences. The Schoolconsiders letters of recommendation, the quality and significance of the candidate's prior workexperience, post-graduate degrees, regional and national leadership, community involvement, andthe quality of their personal statement. Applicants are welcome to discuss how their race, sex,ethnicity, religion, or sexual orientation have affected their lives and any discrimination that theyhave faced. Not all female applicants who were admitted chose to submit a statement along theselines – in fact, most did not. In addition, just over half of the female applicants who wrote such astatement were rejected.

The School has stated a preference to have as much balancebetween men and women as possible and it admits that, in some instances, female applicants areadmitted with lower grades and GRE scores than their male counterparts, but it did not utilize anyfixed numerical quotas or sex balancing.